报告摘要: The fast evolution of computer systems makes it possible to develop and implement very high levels of control and autonomy in vehicles and robotic devices. However, many of these devices are safety-critical, meaning that any malfunction may result in catastrophic consequences, such as loss of human life. It is therefore very important that the software that implements control and autonomy algorithms be of very high quality, and it carry proofs of such high quality. Credible autocoding, a term coined by the author, is about the automatic production of real-time control software that carries its own performance claims, together with proofs supporting them from high-level specifications expressed in Simulink. A prototype credible autocoder developed at Georgia Tech will be discussed and applied to systems, such as an educational device, a jet engine, and an aircraft control system. Future applications to safety-critical robots will be discussed.

报告人信息: Eric Feron has been the Dutton-Ducoffe Professor of Aerospace Software Engineering at the Georgia Institute of Technology since 2005. Prior to that, he was on the faculty of MIT's department of Aeronautics and Astronautics from 1993 until 2005. He holds his BS, MS and PhD degrees from Ecole Polytechnique, France, Ecole Normale Suprieure, France and Stanford University, United States. Eric Feron's interests are to use fundamental concepts of control systems, optimization and computer science to address important problems in aerospace engineering, including: Aerobatic control of unmanned aerial vehicles, multi-agent operations, including air traffic control systems and aerospace software system certification. Eric Feron has published two books and several research papers; his former research students are distributed throughout academia, government and industry. He is an advisor to the Academy of Technologies, France. When he is not in his office, Eric Feron may be found sailing along the coast of Florida.